 The World Health Organization defines malaria elimination as an area where no new indigenous cases have been reported over the past three years. This paper presents a stochastic metapopulation model of malaria transmission which distinguishes between imported, introduced and indigenous cases. It uses data from Zanzibar, Tanzania to parameterize the model and tests the impact of various interventions on malaria incidents. The results show that most new cases on both islands are indigenous, despite high levels of importation. Increasing the coverage of reactive case detection and reactive drug administration can reduce malaria incidents, but for elimination within the next 40 years, transmission reduction in both Zanzibar and mainland Tanzania is necessary. This article was authored by Artre M. Das, Manuel W. Hetzel, Joshua O. Yukich and others.